You are on page 1of 6

2013 Fifth International Conference on Advanced Computing (ICoAC)

An Efficient Delay Aware Multipath GPS


Routing Protocol For Wireless
Sensor Networks
V.P Jaya Chitra, M.R Swnalatha, s. Deepthi
MIT Campus,Anna University
Abstract- There have been several geographic routing voids. The protocol has the following features:
protocols proposed for disseminating data in Wireless Sensor 1) Packet delivery decisions are locally made. No feedback
Networks. In these protocols, routing is based on the location of
from the sink or any knowledge of the global topology is
neighboring nodes and the selection of the next hop is done based
required.
on a direction based or a location based strategy. In this paper, a
novel routing protocol that makes use of this direction based
2) The protocol emphasizes explicitly on fast progress.
strategy to disseminate data in the shortest possible time is 3) Recovery from voids and other failures is done through an
proposed. The protocol uses minimal amount of storage to extremely simple and effective procedure.
determine the next hop in the shortest amount of time possible. 4) The amount of state stored is proportional to the number of
Simulation results obtained using the Castalia simulator show the
neighbor nodes.
efficiency of the proposed protocol over the standard geographic
The rest of the paper is organized as follows. Section-II
protocol GPSR.
presents related work. In Section-III, DAMGPSR is presented.
Index Terms- Delay, geographic routing, wireless sensor Section-IV presents simulation results wherein the protocol is
networks compared with GPSR protocol. The paper is concluded in
Section-V
1. INTRODUCTION
2. RELATED WORKS
The past decade has witnessed a growing interest in sensor
Networks. There have been many applications in growing Mainly all routing protocols of Wireless Sensor Networks
nwnber of domains for WSN. Some of these applications are fall under the ad-hoc category. Generally they are classified
eminently critical and require delay aware mechanisms. Some into Proactive Protocols (table driven) , Reactive Protocols (on
of them are military, health, surveillance, monitoring etc. But demand routing) and Hybrid protocols. Proactive protocols
like DSDV[I], WRP[2], GSR[3], FSR[4] and CGSR[5] are
some of the limited resources like low battery power,
not suitable for Wireless Sensor Networks mainly due to the
relatively low processor capability, small antenna height etc
fact that the limited storage, and energy supply of the nodes
are challenges to be considered while proposing an
are not suitable for storing tables and routes. Hence reactive
mechanism for WSN
protocols like AODV[6], DSR[7] and CBRP[8] are being
There have been routing protocols designed keeping these preferred for WSN. Whereas reactive protocols have the
challenges in mind, but special emphasis has been paid to disadvantage of high delay because route discovery procedure
reduce energy conswnption. This has lead to approaches like has to be initiated every time there is a demand. Also the
tree construction, sleep/awake schedules and probabilistic flooding mechanism that it involves causes a cause of concern
forwarding, all of which can prove to be costly in terms of for the low energy available for WSN.
delay. When delay is taken as the main criterion, there have been
But in contrast to these approaches, geographic routing many approaches to reduce delay. [9] uses DSPA, a protocol
protocols take decisions locally based on the location of their which achieves a very good energy-delay trade-off. But it
neighbors. They are generally robust to topology changes and requires high storage facility as it maintains a routing table.
energy efficient. There are two approaches to geographic [10] proposes DERP. Any node which is interested in an
routing, namely distance based approach and direction based event, constructs a minimum delay tree and uses that for
approach. The former selects the next hop as the neighbor routing. But it uses flooding as part of route maintenance,
closest to the sink, whereas the latter approach selects the node which is not favorable for energy reduction. In [11], an
with the minimwn deviation angle from the line connecting extension of AOMDV( a protocol which chooses multiple
that node and the sink or the required destination node. paths for recovery), called TIDOM is proposed, wherein the
The proposed work is based on the direction based strategy. nodes delay replying to route requests based on the residual
energy capacity. But it has the overhead of running a path
Generally protocols using the distance based strategy fail in
selection algorithm. [12] proposes dividing the network in into
the presence of voids. The usual recovery procedure is the
three zones based on estimated distances and uses different
right hand rule, which is very costly in terms of delay. Hence
forwarding approaches in each zone. It also considers the link
DAMGPSR is proposed which stores a minimal amount of
quality while selecting the route. However this protocol,
state in order to make a quick recovery from failures and though effective assumes that no geographical information of

472
978-1-4799-3448-5/13/$31.00 © 2013 IEEE
2013 Fifth International Conference on Advanced Computing (ICoAC)

the network is available.


QoS factors have been introduced in general routing
protocols, by considering delay as part of the cost in routing.
[13] uses a Delay Index and Connectivity Index in the OLSR
protocol and achieves a lower delay. [14] proposes DMSR.
Here for every route, accumulated delay is calculated for
every path and only the nodes having a delay less than a
threshold continue the process. Of the multiple paths, a path
selection algorithm is used to select one of the paths, and is
also useful during route maintenance. Generally geographic
Fig 1: A GPS built-in sensor node
routing protocols [15]-[17] are being considered for WSN.
They are not affected at all by topology changes due to failure B. Neighbor Table Formation
of nodes as they can simply select some other node as the next By sending periodic hello messages, the neighbor table
hop neighbor. Moreover, they do not require knowledge of the of every node is maintained. Thus if a node does not receive a
entire topology. Hence geographical routing is our preferred packet after a threshold time, it is dropped from the neighbor
approach for reducing delay. list. The neighbor table is of the form shown in Table-I.
Recently a reactive protocol that makes use of the
geographical information has been proposed in [16]. [17] Table.1 Neighbor - Table
proposes an approach wherein the Compass Routing (CR) and x- y- Timestamp
Greedy Routing (GR) indices are combined. Two nodes with Coordinate Coordinate
the lowest direction angle are chosen. Out of them, the one
which is nearer is chosen. Also a routing protocol which has
MAC embedded in it called DGRAM has been proposed in C. Direction Based Greedy Routing
[18]. Whenever a node receives a packet, it checks all its
In our work though, a geographic protocol which makes use neighbors. Of the nodes that are closer to the destination than
of GPS to find out its own co-ordinates and that of the sink is itself, it chooses two nodes that are closest to the destination
proposed. It chooses a forwarding node based on the shortest and stores them in the forwarding table. It obtains the sender
distance from the destination and an alternative node in case address from the intermediate_sender field and sends an ACK
of failures. But once the packet has been forwarded to the sender. It sets the intermediate sender location field to
successfully, the source immediately drops its own copy of the
its own field, the intermediate_destination field as that of the
packet and the receiving node is considered as the source, else
closest neighbor and the alternate_destination field as that of
it sends the packet down the alternative path. It then starts a
the other neighbor. It sends the packet to the closest neighbor
neighbor discovery procedure to get another neighbor. Thus
and saves a copy into its own cache. Once an ACK is received
optimal path with minimum complexity and minimum
memory complexity has been determined efficiently. from the neighbor, it deletes the packet copy that it has.

D. Recovery and Re-transmission


3. DAMGPSR PROPOSAL
If a node receives a packet, but has no neighbors that are
closer to the destination than itself, it checks the
In this section the three phases of the routing protocol are
alternate_destination field set in the packet. If any of its
explained, namely: Neighbor Table Formation, Direction
neighbors are closer than this alternate destination, it sends it
based greedy routing and then Recovery routing. The
to this neighbor. It then sends an ACK to its sender after
assumptions are
setting the flag greedy_mode to false. The sender on receiving
1) The network is static this ACK, deletes its own copy of the packet and removes that
2) The application is sink-based; i.e. all packets particular node from the forwarding table.
generated have the same destination.
But if the node did not have a node closer to the destination
A. GPS Built-in Sensors than the node specified in the alternate destination field, it
Our proposed routing protocol uses the advantage of GPS sends a NACK to the sender. The sender sets the failed flag
built-in sensor as shown in Fig. 1 to minimize the delay in true in the neighbor table and sends it to the alternate
sending the data. The GPS built-in sensor helps us to find the destination.
remote locations of the destinations, so that our algorithm However, if a node has only one node closer to the
effectively uses the data to route packets with minimal delay. destination than itself, it sets the alternate_destination field as
Nowadays GPS enabled sensor nodes are rapidly growing empty. The receIYmg node, if it needs the
and are becoming widely used. GPS enabled sensors have alternate_destination field, but fmds it to be empty, it has to
functions that can return the latitude and longitude of the use the location of the sender. If it fails to find such a
sensor node i.e. their location. The value of latitude and neighbor, it sends NACK to the sender and simply sends the
longitude (N,E) can be used as node ID. With the help of these packet to the neighbor closest to the destination.
(N,E) value, shortest path can effectively be calculated.

473
2013 Fifth International Conference on Advanced Computing (ICoAC)

(LJIiIude.LongitudtHN,E)
DAMGPSR Neighbor Table Formation

Input-none
Output- Each node knows its neighbors and maintains a (12,18)

Neighbour_Table
begin procedure
for all node; in the node[] do
each node; broadcast HELLO packet
all nodej that hears the broadcast message updates the node; Oeslinllion
value in its Neighbour_Table
nodej that is neighbor of node; sends HELLO packet to node; (07)3)

end for
end procedure
(04,21)

(01,04) (01,13)

DAMGPSR Routing

Input- nodes-the source node, Neighbor_Table of nodes


,noded-the destination node, DATA packet Fig. 2: Illustration of DAMGPS routing mechanism stage 1
Output- Forwarding_Table- contains which node is the next forms its Neighbor_Table. Let (08,02) be the source and
forwarding source and another node used as a back up for (07,23) be the destination. The Neighbor_Table of the source
multipath will look as shown in Table.2
begin procedure Similarly, the Forwarding_nodes Table will look like the one
if (noded in the Neighbor_Table) below since the nearest two nodes to the destination are
send DATA packet (06,05) and (11,04) compared to (01,04).
else
for each node; in the Neighbor_table do
calculate distance; � >J(lat_node;-lat_nodecJ2 Table.3: Forwarding Table of(08,02)
+ �on_node;-long_nodecJ2 Latitude Longitude
end for
Forwarding_Table[] � min(distanceJ and 06 05
min2(distanceJ
for each node; in the Forwarding_Table do
11 04
send DUMMYpacket from nodes
end for In the same manner, (06,05) will act as new source and the
send DATA packet to the node; in the same procedure is carried out. The Neighbor_Table and the
Forwarding_Table that replies fust for DUMMYpacket Following_table of (06,05) will be the same since only two
if (no ACK packet after timeout) nodes are present in the Neighbor_Table.
/* for the case of void or node mobility */
choose next node from the Forwarding_Table
send DATA packet
end procedure

(12,18)
E. Illustration of The Routing Procedure
The routing procedure of DAMGPS is illustrated in Fig.2 for
stage-I, in Fig.3 for stage-2 and in Fig.4 for stage-3. Each
node broadcasts HELLO messages to every other node and

Table.2 : Neighbor Table of (08,02) (07,23)

Latitude Longitude

(04)1)
06 05
(01,04) (01,13)
11 04
01 04
Fig 3: Illustration of DAMGPS routing stage:2

474
2013 Fifth International Conference on Advanced Computing (ICoAC)

Table.4: Neighbor Table and Forwarding table of(06,05) control packet are and respectively 64 bits each. The packet
Latitude Longitude generation rate is 5 packets/sec and radio range is as far as 30
metres.
05 09 Table 5:Parameter list of simulation
10 lO Parameter Value
Average node degree 4.59 5.16 8.02 9.65 11.91
If there occurs some error and if node (08,02) did not receive 15.82 20.48
any acknowledgement after certain timeout, the algorithm will Network size 150
go along the next chosen node in the Following-Table, Data packet size 64 bit
(11,04). Network topology Random
This process continues until the destination is reached. Mobility Static
When node (08,20) is reached, the Neighbor_Table is checked
and the algorithm finds that the destination is the next best The end-to-end delay performance of DAMGPSR is shown
node to route the data. Hence the procedure terminates after in Fig.5. The proposed protocol offers a significantly better
the DATA packet is sent. performance in delay efficiency, especially in presence of
Thus at each stage, the packet gets geographically closer to voids and node failures. This is because unlike GPSR, the
the destination and from the closest path the quickest one is proposed protocol stores some path information before packet
chosen. Hence the data is routed along the quickest path and forwarding.
always the data moves one step closer to the destination. The
multipath helps in recovering from failure in case of void or
mobility.
gpsr­
damgpsr -+-
0.09

(12.18)
0.08

0.07

0.06

(07�) 0.05

0.04
(04)1)

(01.13) 0.03 '----'-_--'-_....1...-_.1..----'_--'-_-'-_-'---'


4
U71l1,o.ICll54

Fig. 5 : Comparison between DAMGPSR and GPSR


Fig. 4: Illustration of DAMGPSR routing stage:3
DAMGPSR works better than GPSR especially when the
networks are less dense.
4. SIMULATION RESULTS
B. Packet Formats Used in Simulation
DAMGPSR and GPSR have been implemented using the
Castalia framework over the OMNeT++ simulator in order to The packet formats used in simulation are listed in Table.6.
systematically compare the performance of DAMGPSR along The "kind" field specifies the type of the routing packet,
with the geographic protocol GPSR. whether data, ACK or NACK packet. The data, ACK, NACK
packets are specified by values 0, 1, 2 respectively.
A. Implementation Results In Table.7, the additional fields required for data packet are
listed. The Intermediate_Source field specifies the forwarding
We evaluate the performance on different topologies with node from which the node originated. This is especially useful
different node degree i.e. the average number of neighbors for when ACK packets are received, because the node can
a node. The parameters used for the simulation have been identify easily which packet it has to delete from its array.
depicted in Table-I. The nodes are initially placed in a random Similarly the Intermediate_Destination field specifies the next
manner in a rectangular region. The size of the data packet and forwarding node to which the packet has to be sent. The

475
2013 Fifth International Conference on Advanced Computing (ICoAC)

Table.6 : Packet format used in simulation memory requirements which is desirable for Wireless Sensor
Type Networks in general. The protocol also minimizes flooding of
Field
packets and ensures that the number of copies of the packet
Double held by the sensor networks is minimum. This is highly
RSSI
desirable for the energy constrained Wireless Sensor
Double Networks. The simulation results show the lower delay
LQI
required by DAMGPSR over a standard geographic protocol
Int like GPSR.
nextHop
Int REFERENCES
lastHop
[1] C. Perkins and P. Bhagwat. "Highly dynamic destination­
String
Source sequenced distance vector routing (dsdv) for mobile computers."
ACM SIGCOMM: Computer Communications Review, pp:234-244,
String October 1994.
Destination
Int [2] S. Murthy and 1. Garcia-Luna-Aceves, "An efficient routing
Sequence No.
protocol for wireless networks. ACM Mobile Networks and App. 1.,
Int Special Issue on Routing in MobileCommunication Networks, pp
Kind
183-197, October 1996.

time_delivery field is used to record the time at which the [3] T.-W. Chen and M. Gerla, "Global state routing: A new routing
packet originated so that the delay experienced by the packet scheme for ad-hoc wireless networks." In Proceedings of IEEE
can be easily calculated. ICC'98 on pp 171-175 Vol-I, 1998.

[4] A. Iwata, C.-c. Chiang, G. Pei, M. Gerla, and T.-W. Chen.


Table.7 : Additional Fields for Data Packet
"Scalable routing strategies for ad hoc wireless networks." IEEE
Field Type Journal on Selected Areas in Communications, Special Issue on Ad­
Application ID Int Hoc Networks, pp 1369-1379, August 1999.
Data SequenceNumber Int
Data Double [5] c.-c. Chiang, "Routing in clustered multihop, mobile wireless
Source xCoordinate Int networks with fading channel." In Proceedings of . IEEE SICON'97,
pp 197-211, April 1997.
Source yCoordinate Int
Destination xCoordinate Int [6] C. E. Perkins and E. M. Royer. "Ad hoc on-demand distance
Destination yCoordinate Int vector routing." In Proceedings of the 2nd IEEE Workshop on
Intermediate Source String Mobile Computing Systems and Applications, pp 90-100, New
Intermediate Destination String Orleans, LA, February 1999.
Time Delivery Sim time
[7] David B. Johnson , David A. Maltz , Josh Broch, DSR: the
dynamic source routing protocol for multihop wireless ad hoc
C. Advantages OJDAMGPSR networks, Ad hoc networking, Addison-Wesley Longman Publishing
Co., Inc., Boston, MA, 2001.
No implosion- even though the DAMGPSR is multipath
algorithm, implosion never occurs because, unlike other [8] M. Jiang, J. Li, and Y. Tay, "Cluster based routing protocol",
multipath algorithms, duplicate copies of the same data is not August 1999. IETF Draft.
sent unless the first chosen path fails or becomes void or
unavailable. Hence the duplication of data to the same node is [9] Haojun Huang, Guangmin Hu, Fucai Yu, " Delay-Sensitive and
avoided. Power-Aware Routing In Wireless Ad Hoc Networks"
Communication Technology (ICCT), 12th IEEE International
Delay Efficient- Since at each step we are geographically one
Conference pp 496 - 499 2010 IEEE.
step closer to the destination and the failure does not
reinitialize the process of route fmding because of the [10] Liming He,"Delay-Minimum Energy-Aware Routing Protocol
multipath, it is more delay efficient in emergency applications. (DERP) for Wireless Sensor Networks", Software Engineering,
Memory Efficient- It is highly memory efficient because there Artificial Intelligence, Networking, and Parallel/Distributed
is no need to store entire routes nor many copies of the packet. Computing, Eighth ACIS International Conference pp: 155 - 160Aug
Less Complexity- Both the routing and failure recovery are 2007.
extremely simple. For every destination, Euclidean distances
need to computed for all neighbours only once. Extensive [11] Woncheol Daeyoung Kim, Taehong Kim, Seong Hoon
Kim,"Time Delay On-demand Multipath Routing Protocol in Mobile
computations are not necessary.
Ad-hoc Networks" Ubiquitous and Future Networks (ICUFN), 2011
Third International Conference pp 55 - 60 June 2011.
6. CONCLUSION

DAMGPSR, a new routing protocol that reduces delay, has [12] Xin Ming Zhang, En Bo Wang, Jing Jing Xia, and Dan Keun
been proposed in this paper. The protocol has very low Sung, "An Estimated Distance-Based Routing Protocol for Mobile
Ad hoc Networks" IEEE TRANSACTIONS ON VEHICULAR

476
2013 Fifth International Conference on Advanced Computing (ICoAC)

TECHNOLOGY, VOL. 60, NO. 7, pp 3473 - 3484 SEPTEMBER [16] Rong Ding, Lei Yang, "A Reactive Geographic Routing
2011. Protocol for Wireless Sensor Networks" Intelligent Sensors, Sensor
Networks and Information Processing (ISSNIP), pp: 31 - 36 Dec
[13] Kunagorn Kunavut* and Teerapat Sanguankotchakorn , "QoS­ 2010.
aware Routing for Mobile Ad hoc Networks Based on Multiple
Metrics:Connectivity Index (CI) and Delay" Electrical [17] Alok Kumar and Shirshu Varma, "Geographic Node-Disjoint
Engineering/Electronics Computer Telecommunications and Path Routing for Wireless Sensor Networks" IEEE SENSORS
Information Technology (ECTI-CON), pp: 46 - 50 May 2010. JOURNAL, VOL. 10, NO. 6, pp 1138-1139 JUNE 2010.

[14] Shun Liu Jian Liu,"Delay-aware Multipath Source Routing [18] DGRAM: A Delay Guaranteed Routing and MAC Protocol for
Protocol to Providing QoS Support for Wireless Ad Hoc Networks" Wireless Sensor Networks, IEEE Transactions On Mobile
Communication Technology (ICCT), 2010 12th IEEE International Computing, VOL. 9, NO. 10, pp 1407-1423 OCTOBER 2010.
Conference pp 1340 - 1343 2010 IEEE.

[15] Brad Karp , H. T. Kung, GPSR: greedy perimeter stateless


routing for wireless networks, Proceedings of the 6th annual
international conference on Mobile computing and networking,
p.243-254, August 06-11, 2000, Boston, Massachusetts, United
States.

477

You might also like